Qualitative And Quantitative Analysis In Animal Fibers By Visible And Near Infrared Reflectance Spectroscopy

Animal fiber is one of the important products in the animal husbandry and thehigh-grade textile materials in the textile industry, but their textile products havedifferent performance and market price, so that adulteration phenomenon has beenemerged on the market, and has hindered development of the textile industry. NIRshas advantages such as good speediness, accuracy, repeatability. so it is widely used inthe animal husbandry, food, medicine, chemical industry and other fields. In order toprovide technical reference for the commercialization processed in the animalhusbandry and textile industry, using the Vis-NIRs to identify the similar animal fibersand measure the content of the blend animal fibers in the Gansu province.The animal fibers were collected from different areas of the Gansu province thatwere homogeneous sheep-wool, sheep cashmere, cashmere and camel hair. It wasused to research on visible-near infrared spectral absorption characteristic, computethe GH and NH to simplify the spectral data and exclude exceptional spectrum,analyze by principal component-mahalanobis distance pattern, establish thequalitative model for the similar animal fibers and quantitative analysis model forcashmere content in the blend samples by visible and near infrared reflectancespectroscopy(Vis-NIRs).The average spectrum of the different fibers(homogeneous sheep-wool, sheepcashmere, cashmere, camel hair) were observed that showed the same absorptioncharacteristics and made the same absorption peak position at1736nm,1940nm,2054nm,2176nm,2328nm,2354nm in near infrared wavelengths(1100-2500nm).The principal component-mahalanobis distance pattern with the bestpretreatment and principal component contribution rate, it was used to discriminatethe different similar samples within group, there were three groups that includedhomogeneous sheep-wool and cashmere; sheep cashmere and cashmere; sheep wool,cashmere and camel hair. The discriminating boundary between different categorieswithin group were clear in the3D graph of the best order principal component scores.The principal component regression(PCR) was used to establish the calibration qualitative model about three groups. The samples of homogeneous sheep-wool (n=65)and cashmere(n=65) were established the calibration qualitative model, the result ofthe external validation was correct completely. The samples of sheep cashmere(n=32)and cashmere(n=32) were established the calibration qualitative model, the predictedresult of the calibration set and the validation set was97.5%、100%, respectively. Thesamples of sheep wool(n=83), cashmere(n=83) and camel hair(n=60) were establishedthe calibration qualitative model that the predicted result of the calibration set and thevalidation set was98.8%、98.4%, respectively. The results were greater than97%, thecalibration qualitative model of PCR can be realized to identify the similar animalfibers by Vis-NIRs.The fine sheep wool and cashmere were mixed the blend samples(n=92), theywere analyzed on qualitatively and quantitatively by Vis-NIRs between the puresamples(n=30). The spectrum of the blend samples showed that the same absorptionpeak position with the pure samples, the absorption value of same absorption peakposition were decrease with increasing content cashmere of the blend. The identifiedboundary between two categories was clear by the3D graph of principalcomponent-mahalanobis distance pattern; The calibration set of qualitative model wasused to establish calibration qualitative model, the result of the external validation wascorrect completely; Quantitative analysis models for cashmere content in the blendsamples were developed with modified partial least squares regression, and combinedwith the pretreatment of WMSC and2nd derivative, the six of best principal factor.The values of R squared (RSQ) and standard error of the calibration(SEC) were0.9900and0.0228in the internal validation of the model, respectively; the ratio ofperformance to standard deviate(RPD) was more than5in the external validation ofthe model, T test of the actual and predicted value were no significant difference onboth sides. Results from this experiment indicate that Vis-NIRs can be realized toidentify and quantify about the content of the blend animal fibers.